Wednesday, January 23, 2019

How valuable was Mariano Rivera?

Yesterday, Mariano Rivera became the first player ever to be elected to the Hall of  Fame unanimously, and on his first try to boot--something that players like Babe Ruth, Ted Williams, Willie Mays, Henry Aaron, Mike Schmidt, and many others, failed to achieve.  He was elected because he was (as we shall see) a very effective pitcher, who played what many people believe to be a key role for the New York Yankees for many years, during most of which they enjoyed extraordinary success.  He certainly proved, over a long period of time, that he had Hall of Fame ability.  Yet for reasons having nothing to do with his ability, he actually contributed extraordinarily few extra wins to his team over the course of his long career.

The principle of reductio ad absurdum has solved a number of problems in mathematics, both theoretical and applied.  Essentially, it suggests that if the logical extension of a premise leads to an absurd conclusion, the premise itself is false.  For several decades now fans and writers have believed, and baseball organizations have acted as if, a pitcher who can reliably protect a lead of 1-3 runs for one inning (the ninth) plays a critical role in his team's success.  In 2016, in a typically thorough and insightful presentation at the SABR Convention in Miami, Dave Smith of retrosheet.org proved that that assumption is false.  An analysis of the career of Rivera--generally regarded as the greatest closer of all time--will confirm that conclusion in spades.

After a year as the Yankees' set-up man in 1996--a year to which we shall return--Rivera was the Yankees' closer for 16 of the next 17 years.  He was an extraordinarily effective pitcher for most of those seasons.  In Baseball Greatness, I use the measurement of Wins Above Average to measure the value of all players, and by that measure, Rivera appears to be the best relief pitcher of all time. Only once did he ever come close to the 4 Wins Above Average figure that represents a superstar season, but no other reliever has ever pitched enough to reach that figure either.  He exceeded 2 WAA in four out of five seasons from 2003 through 2008--seasons in which he pitched between 70 and 80 innings.  No other reliever--including those from earlier eras who pitched well over 100 innings a year--has been over 2 WAA that many times.  That means that if he could have pitched 220 innings, let's say, at the same rate of effectiveness, he would  have earned about 6.5 or 7 extra wins for his team, which represents an outstanding season for any of the greatest pitchers in history, such as Lefty Grove or Bob Feller or Roger Clemens.  I have no doubt that, given the chance, and if he could have stayed healthy for long enough, Mariano Rivera might have pitched his way into the Hall of Fame as a starter.  But we will never know.

Yet the New York Yankees, as it turns out, got remarkably little value, in terms of games won, out of this great pitcher.  The reason is that, as a closer, Rivera only came into the game when the Yankees already had an overwhelming chance of winning it.

I have not been able to find a thorough year-to-year record of Rivera's saves and blown saves, but an article by Jim Caple of ESPN written on the eve of his retirement has good aggregate data.  Rivera, he reports based on data from Elias, had 652 saves and 80 blown saves. That is a percentage of 89.1.   Unfortunately, I don't know how many of the blown saves resulted in  Yankee defeats.  They obviously lost all the games on the road in which Rivera actually surrendered enough runs to lose the game, and the vast majority--around 90%--of the games in which he fell behind in the top of the ninth at Yankee Stadium.  They should have won about half the games in which he allowed the other team to tie the game.  I'm going to be generous, for reasons that will become apparent, and guess that the Yankees might have won 20 games in which Rivera had blown the save.  That would mean that the Yankees won 92% of the games in which Rivera came into the game in a save situation.

How would an average pitcher have done?

Dave Smith's presentation provides some answers.  Team winning percentages in games in which the team enjoyed a 1-, 2- or 3-run lead going into the ninth inning are extraordinarily consistent from 1912 to 2015.  A 3-run lead gives a winning percentage of 97%, a two-run lead 93%, and a one-run lead,  84%.

Now the Caple article does break down Rivera's saves into 1-run, 2-run, 3-run and 4-run situations, but it doesn't do the same, alas, for his blown saves.  (Caple states, in effect, that even an entrance with a 4-run lead can result in a save, while Smith did not.  I don't know where this discrepancy came from, but I'll count 3- and 4-run leads together for the next calculation.)  I'm going to assume, generously I think, that Rivera blew 2/3 of his blown saves with a lead of just 1 run, and 1/3 with a two-run lead.  Adding that estimate into Elias's figures for successful saves, we would find Rivera entering with a one-run lead 263 times, with a two-run lead 241 times, and with a 3- or 4- run lead, 226 times.  Applying Smith's figures for success in those situations, we find that the Yankees with average 9th-inning pitching could have expected to win 664 of the of the 732 games that Rivera entered in a save situation, or 91% of them.  With  Rivera, we have estimated that they actually won 672 of them, or 92%.  That's 8 extra games in 16 full-time seasons as closer--exactly one-half a game a year.

Now as I have said, Rivera actually exceed 2 WAA a year four times as a closer--the most of any reliever--and he also exceeded 1 WAA on ten other occasions, implying that he was worth more than half a win a year to the New York Yankees.   But he wasn't, because the Yankees used him as a closer.  They used him in games in which the rest of the team--the hitter/fielders, starting pitchers, and other relievers--had already established a win probability of 91%.  There was no way that he could improve on that probability very much, given the way that he was used.  The same is true, obviously, for very other talented pitcher who is wasted as a closer.  Rivera's most valuable year for the Yankees, ironically, w 1996, when he pitched 107 2/3 innings as a set-up man and earned 3.4 WAA, more than he ever did as a closer.

There is, alas, another question that we really need to ask about Rivera and other closers.  Rivera's best seasons as a closer, when he earned more than 2 WAA in four years out of five, occurred at ages 33-38.  It certainly seems reasonable to assume that he managed to sustain that level of effectiveness at that age in part because he pitched only 70 innings a year.  What we have shown, though, is that that reduced his value to his team to a very marginal level.  Rivera was a very fine pitcher, in the very few innings that he pitched, for a very long time.  But based on the actual number of wins that he and other closers contribute to their teams, none of them should even be considered for the Hall of Fame.





Tuesday, January 1, 2019

The election of Harold Baines to the Hall of Fame

The selection of Harold Baines to the Hall of Fame has excited a great deal of comment, most of it negative, because so few people, apparently, have ever regarded him as a potential Hall of Famer.  Once some one is in, of course, they are in forever, and there's no point ranting and raving about any mistakes that voters have made.  Yet it is worthwhile, it seems to me, to ask a few questions that situate this choice within the pantheon of the Hall, and help place Baines accurately among baseball's greatest players.  I shall focus on three questions:

1.  How exactly was Baines selected?

2.  How does Baines compare to broadly similar players already in the Hall?

3.  How does Baines compare to the other players who were on the ballot this year?

Baines was chosen by one of the subcomittees, one might say, of the old Veterans Committee.  Called  the Today's Game Committee, it "considers retired Major League Baseball players no longer eligible for election by the Baseball Writers' Association of America (BBWAA), along with managers, umpires and executives, whose greatest contributions to the game were realized from the 1988-2016 era."   The committee membership included  Hall of Fame members Roberto Alomar, Bert Blyleven, Pat Gillick, Tony La Russa, Greg Maddux, Joe Morgan, John Schuerholz, Ozzie Smith and Joe Torre; major league executives Al Avila (Tigers), Paul Beeston (Blue Jays), Andy MacPhail (Phillies) and Jerry Reinsdorf (White Sox); and veteran media members/historians Steve Hirdt, Tim Kurkjian and Claire Smith.  The other players on the ballot were Albert Belle, Joe Carter, Will Clark, Orel Hershiser, and Lee Smith. Smith also won election.

2.  I am going to compare Baines's record to others using the method I developed in my book, Baseball Greatness.  That method ranks players by the number of seasons of 4 or more Wins Above Average they had during their careers.  For hitters like Baines, WAA is computed using the offensive numbers from baseball-reference.com, but substituting Michael Humphreys's DRA fielding statistics for the ones baseball-reference uses, and eliminating position adjustments, which tend to be arbitrary.  These differences of approach help Baines.  DRA shows him to have been a significantly better fielder than baseball-reference does in the early years of his career when he played the outfield, and baseball-reference routinely subtracts about 10 runs above average for DHs, which of course was what Baines was for most of his career.  4 WAA is my definition of a superstar season because it turns out that it defines how good you have to be to be the MVP on a pennant-winning team.  The vast majority of pennant winning teams have had at least one player that good.

Now the number of seasons of 4 WAA or more that a player has does correlate reasonably well with lifetime totals, since a player who is great enough to have five such seasons (stand by for more explanation) will generally have a long career including a number of seasons in the 2-3 WAA range, as well.  As it turns out, 5 seasons of 4 WAA or more has in practice been the definition of a solid Hall of Famer, according to the actual voting results.  Of the 29 players with exactly five such seasons, 19 of them--Phil Niekro, Cy Young (1901 and later), Stanley Covaleski, Rod Carew, Jim Thome, Dazzy Vance, Harry Heilmann, Goose Goslin, Charlie Gehringer, Joe Gordon, Jackie Robinson, Hal Newhouser, Joe Morgan, George Brett, Jim Rice, Tim Raines, Ryne Sandberg,  Harry Hooper, and Tom Glavine--19 players who are in the Hall of Fame.  Three much older players from this group--Charlie Keller, Gil Hodges, and Wes Ferrell--are not in.  The remaining 7 players with 5 WAA are Roy Halladay, Curt Schilling, Larry Walker, Mike Mussina, and Jim Edmonds, who are on this year's BWAA ballot, and David Ortiz and Clayton Kershaw, who are not yet eligible.  On the other hand, of the 51 players in history with exactly four seasons of 4 WAA or more, only 17 of them are in the Hall of Fame--and many of them are pitchers, for whom the actual standards, by this method, tend to be a good deal lower.

There are, of course, a good many players in the Hall of Fame who do not have even 4 seasons of 4 WAA or more, but the vast majority of them are catchers or middle infielders, and they are not, therefore, good comparisons to Harold Baines, who began his career as an outfielder and spent most of it as a DH.

By this method, Harold Baines has no claim to the Hall of Fame, for the simple reason that he never, in his whole career, had a season of 4 WAA or more.  Not one.  Here are his annual WAA totals (I have combined the individual team totals for the several years in which Baines played for more than one team.)

1980     -2.1 WAA
1981       0.9 WAA (strike-shortened season)
1982       0.8 WAA
1983      1.3
1984      3.5
1985      2.2
1986      2.4
1987      0.5
1988      0.6
1989      2.5
1990      1.3
1991      2.1
1992     -0.3
1993      1.8
1994      0.1
1995      1.8
1996      2.6
1997      1.1
1998      0.3
1999      1.9
2000     -0.8
2001     -1.3

In 21 years, Harold Baines never had a superstar season.  He had 6 seasons of 2.0-3.9 WAA, all but one under 3 WAA, which I define in my book as "star seasons."  He had five seasons of 1-1.9 WAA, making him a modest asset on a winning team, six seasons in which he was essentially average (from -0.9 to +0.9), and his last season was below average.  For an outfielder/DH this is a very poor record for a Hall of Famer.

Are there any outfielders in the Hall whose records are very comparable to Baines'? Yes, at least three.  One, indeed, is indisputably worse than Baines, and ranks as a candidate for the weakest non-pitcher in the Hall of Fame.  That is Lloyd Waner, who got lots of base hits, played in a high-offense era, and benefited from the glow emitted by his brother Paul, a genuine all-time great.  Lloyd Waner also had a long career (18 seasons to Baines' 22) without ever having a season of 4 WAA or more, but his record is much weaker than Baines's.  He topped 2 WAA only once (in 1931) and 1 WAA only four times.  He was an average player, or worse, for nearly his whole career.

It is interesting to compare Baines to some of the other weaker outfielders who have reached Cooperstown.  Those most similar to Baines in terms of the length of their careers are Sam Rice (2404 games to Baines's  2830), Max Carey (2476 games), Zack Wheat (2410 games), and Heinie Manush (2008 games.) Rice was elected for a reason that apparently played a big role in Baines's selection as well: both of them finished their careers with close to 3000 hits, which has traditionally meant certain inclusion in the Hall of Fame.  As it happens, Rice was also rather unlucky insofar as he did not reach the majors until he was 25 and had his first full season at 27.  Nonetheless he played in 20 seasons, like Baines, but he was an outstanding center fielder, as it turns out, and he had much more impact on his teams than Baines did.  He had one superstar season (5.6 WAA in 1920, when he saved a remarkable 37 runs in center field.)  He was over 3 WAA five more times (compared to once for Baines), and over 2 twice more.  He was a star, although not the MVP, on the Washington Senators' pennant winning teams of 1924-5.  He is one of the weaker outfielders in the Hall, but he was significantly better than Baines was.  Max Carey had three superstar seasons of 4 WAA or more and three more seasons over 3 WAA (he too was an outstanding outfielder.)  Zack Wheat had three seasons in which he topped 5 WAA, and four seasons over 3 WAA.  Manush's two best seasons are better than Baines's two best (4.7 and 3.5 WAA), but Baines's record is better than his after that.  Outfielders with much shorter careers whose status as Hall of Famers is questionable include Chick Hafey, Ross Youngs, Earl Averill, Chuck Klein, and Hack Wilson.  Hafey is the only one of that group without a superstar season and he was definitely a less valuable player than Baines.  Youngs had two seasons over 4 WAA and three more over 2; Averill topped 4 WAA just once, and 3 WAA three times (his fielding hurt him badly; and Klein had three superstar seasons. Wilson, another fielder, never topped 4 WAA for a season, even in 1930, when he hit 56 homers and drove in 191 runs but cost the Cubs -33 runs in the field. He was over 3 WAA four times, however.  Combs is one of many players who were elected to the Hall of Fame thanks to their teammates.  He had a relatively short career with the Yankees--12 years, including 3 with less than 100 games--and he topped 4 WAA only once, in 1927, with 6, thanks largely to by far his best season ever in the field.  He had three star seasons, and thus, with the exception of that one big year, he was not as good as Harold Baines.  One could therefore argue that Baines ranks above Lloyd Waner, Chick Hafey,and Heinie Manush among outfielders in the Hall, but that's about it.

Baines is, I believe, the first selection to the Hall who spent the vast majority of his career as a DH.  Two other similar candidates are Edgar Martinez, who is on the ballot this year, and David Ortiz, who is not yet eligible.  Both are in a completely different league than Baines.  Martinez had an outstanding 7 seasons with 5 WAA or more, making him statistically overqualified for the Hall (although his career path, as I mentioned in an earlier post, raises some questions), and Ortiz finished his career with 5 such seasons.  Indeed, it is an interesting question as to why, in the high-offense era of the 1990s, several AL teams were willing to play Baines as their regular DH even though he was contributing only marginally to their success.

3.   Among the players on this year's ballot, Albert Belle had much better Hall of Fame credentials than Baines did.  Belle played only 10 regular seasons but he performed at a superstar level in 4 of them.  That is not enough, as we have seen, to give him a better than 50% chance of reaching the Hall, but Belle was the MVP on the Cleveland Indians' dynasty of the late 1990s for several years. Lee Smith was elected as a closer--but no closer, including Mariano Rivera, has ever had 4 WAA even once. They do not pitch enough to have that much impact.

The old Veterans Committee was responsible for most of the worst selections to the Hall of Fame, and the new committees look like they will do just as badly.  They have failed to elect the players that the  BWAA  unfortunately ignored despite ample credentials, such as Charlie Keller (whose great career was shortened by injury), Gil Hodges, and Keith Hernandez, while adding some players whose credentials are weak.   They do not seem to be in the least interested in any modern sabermetric measurements.  But in an era where rationality is under attack on many fronts, this is not surprising.